--On Thursday, August 22, 2002 12:51:15 -0500 Tony Shadwick <tshadwick AT
stgroupllc DOT com> wrote:
I'd like to say up front that I got sattled with a FreeBSD based ISP with 19
servers that are totally undocumented. Thank you very much. :P
Now that I've got that out in the open, I have one of these systems with a
SCSI Quantum DLT 4500 tape changer attached that takes backups from all the
systems, including itself. These backups have, apparently been working
flawlessly for quite some time.
One problem: They never tested to see if they could recover apparently, as
I cannot. :\
From the clients, if I type
amrecover -C daily -s onyx.ourdomain.com -t /dev/nsa0
I get:
AMRECOVER Version 2.4.1p1. Contacting server on onyx.ourdomain.com ...
amrecover: Error connecting to server: Operation timed out
If I allow it to sit there there long enough, it will eventually time out.
Okay, so for one reason or another the server daemon is unreachable.
Since you can connect locally, perhaps you are running ipchains/iptables,
tcpwrappers, PAM, or something else that is not allowing outside connections.
Let's
try recovering from onyx, to onyx.
amrecover -C daily -s localhost -t /dev/nsa0
As someone already pointed out, the -t is tapeserver not device.
onyx# amrecover -C daily -s localhost -t /dev/nsa0
AMRECOVER Version 2.4.2p2. Contacting server on localhost ...
220 onyx AMANDA index server (2.4.2p2) ready.
200 Access OK
Setting restore date to today (2002-08-22)
200 Working date set to 2002-08-22.
200 Config set to daily.
501 No index records for host: onyx.ourdomain.com. Invalid?
Trying onyx.ourdomain.com ...
501 No index records for host: onyx.ourdomain.com. Invalid?
amrecover>
Errr....okay, so it lets me in, but claims that there are no index records.
Do your dumptypes contain 'index yes' ? I think by default indexing is
turned off.
I have to admin to being baffled. But wait, there's more...
reboot
tail /var/log/dmesg.today
Aug 21 15:49:51 onyx inetd[212]: amidxtape/tcp: bind: Address already in use
Aug 21 15:49:51 onyx inetd[212]: amandaidx/tcp: bind: Address already in use
Aug 21 15:49:51 onyx inetd[212]: amanda/udp: bind: Address already in use
Aug 21 15:51:28 onyx inetd[212]: amanda/udp: bind: Address already in use
Aug 21 15:51:28 onyx inetd[212]: amandaidx/tcp: bind: Address already in use
Aug 21 15:51:28 onyx inetd[212]: amidxtape/tcp: bind: Address already in use
Eh...okay? Address already in use?
So...this is kinda freaky. I took a look at /etc/inetd.conf, and everything
looks good there. Obviously something in inetd is choking, and even
stranger is that I'm not completely certain BIND is even running on onyx.
Really weird. Any ideas?
The error isn't because of the name resolver BIND, it is from inetd telling you
that it received a connection request on those ports but couldn't start the
specified program associated with that port since some program was already
bound to [in control of] those ports. Sometimes the errors result from too
many requests and the programs not releasing the port fast enough between
requests.
Frank
Tony Shadwick
Manager of Internet Services
Strategic Technology Group
314-480-1324
--
Frank Smith fsmith AT hoovers
DOT com
Systems Administrator Voice: 512-374-4673
Hoover's Online Fax: 512-374-4501
| 